Amyloid plaque buildup in the brain is one of the hallmarks of Alzheimer's disease. But one perplexing aspect of amyloid pathology, known for decades from autopsy findings (see Crystal et al., 1988), is that many cognitively normal people are walking around with heads full of amyloid and seem to suffer no ill effects. The advent of imaging technologies that illuminate amyloid deposition among the living illustrates the point even more. The finding begs the question: Are plaques themselves actually harmful?

Two recent longitudinal studies probed that question in non-demented individuals. One, by Michael Weiner and colleagues at the University of California, San Francisco, and published in the October 29 Cerebral Cortex, found in an elderly cohort that high brain Aβ levels did correlate with cognitive decline, brain atrophy, and metabolic dysfunction, but only among people who already had MCI or AD. In healthy controls, high brain Aβ levels correlated only with atrophy in specific subregions of the brain. A second paper by Anders Dale, University of California, San Diego, and colleagues, published in the October Annals of Neurology, reported that CSF Aβ42 in cognitively normal individuals correlated with brain atrophy only when phosphorylated tau was elevated in cerebral spinal fluid (CSF). Such findings could have implications both for predicting disease progression and picking apart Alzheimer's etiology.

"This research is important not only for identifying biomarkers which have the most clinical potential, but also for better elucidating the pathological processes that are taking place during the natural course of the disease," wrote Anne Fagan, Washington University School of Medicine in St. Louis, Missouri, in an e-mail to ARF (see full comment below).

In the Weiner paper, first author Michael Ewers and his team analyzed two-year data from 465 elderly adults recruited into the North American multicenter Alzheimer’s Disease Neuroimaging Initiative (ADNI). Some were healthy controls, while others were diagnosed with either amnestic mild cognitive impairment (MCI) or probable AD. Within each of those three groups, people were categorized as having "high" and "low" brain Aβ levels, based on whether they fell above or below a certain threshold in a previously obtained positron emission tomography (PET) scan with Pittsburgh Compound B (PIB). ADNI investigators conducted a series of tests: the Alzheimer's Disease Assessment Scale (cognitive section) and the delayed recall portion of the Rey Auditory Verbal Learning Test to measure cognition; fludeoxyglucose (FDG)-PET to assess brain metabolism; and magnetic resonance imaging (MRI) to determine gray matter volume.

As might have been expected for the MCI group, high brain Aβ associated with faster decline in cognitive ability, lower brain volume within temporal and parietal brain regions, and reduced parietal metabolism. In healthy controls, high brain Aβ levels correlated only with faster loss of medial temporal lobe and precuneus volume. There was no association between Aβ and either cognitive decline or metabolism in this group. Previously, coauthor Paul Aisen, University of California, San Diego, reported at conferences that PIB-positive normals appear to change a bit faster than PIB-negative people on the Mini-Mental State Exam and ventricular volume over the course of two years (see ARF Webinar and ARF related conference story).

"Even though the levels of Aβ in healthy controls can be high, we don't see a rate of decline anywhere close to that observed in MCI patients with similarly high levels of Aβ," said Ewers. "That came as a surprise."

Why do these people with high levels of Aβ in their brains show no manifestations of disease? Several possibilities could explain the discrepancy, wrote the authors. Healthy controls with elevated Aβ levels could be more resistant to Aβ pathology. They could have other factors that mitigate Aβ's neurotoxic effects. Or perhaps these brains have been accumulating Aβ for a shorter time period and cognitive decline is simply further down the road than the two-year follow-up available to this study.

If the authors had followed this cohort for a longer period of time, they most likely would have seen decline in healthy controls, said Prashanthi Vemuri of the Mayo Clinic in Rochester, Minnesota, who was not involved with the research. "Two years might be too short a time," she said. She believes that the findings fit in nicely with the amyloid cascade hypothesis, and that the healthy controls in the study just had not yet progressed to later stages of the disease.

But Scott Small, Columbia University, New York, said that, with the sensitive and reliable metrics used in this study, two years should have been long enough to pick up a shift if the healthy controls with high levels of Aβ plaques were headed for disease. Some would argue that the data mean other factors mitigate Aβ's affects, he said, and the Dale paper (see below) suggests that it is phosphorylated tau. "I think there's no doubt that amyloid is necessary for AD, but it's not perfectly clear yet if it's sufficient," Small told ARF. Though this paper does not definitively answer the question of whether Aβ is causative, he predicted that studies like this one will soon resolve the fundamental question of Aβ's action. "If these new studies start suggesting that amyloid can happen as part of normal aging, and what really causes Alzheimer's is an interaction between Aβ and tau, and tau pathology is caused by something else, then the amyloid hypothesis might need to be modified." Not only would that inform drug trial design and explain why amyloid reduction is sufficient for treatment, but it would also change the field's understanding of the disease, he said.

The second paper by Dale and colleagues proposes that it is phosphorylated tau that is needed for Aβ to exert its toxic effects. First author Rahul Desikan, University of California, San Diego, and his team looked at cerebral spinal fluid biomarkers and longitudinal MRI scans of 286 non-demented older adults from ADNI—some healthy older controls and others with mild cognitive impairment. Brain atrophy, most notably in the entorhinal cortex, correlated with low CSF Aβ42 (which corresponds to high amyloid plaque levels in the brain) only when phosphorylated tau levels were high in the CSF as well. Neither Aβ42 nor phosphorylated tau by itself correlated significantly with atrophy.

Paired with findings in animal models that suggest a mechanistic link between tau and Aβ (see ARF related news story on Ittner et al., 2010 and ARF related news story on Roberson et al., 2007), the data support the idea that "tau is playing a critical role in modulating the relationship between Aβ and synaptotoxicity," Desikan said. Not only does this have implications for the cause of Alzheimer's disease, but it also means that clinicians should screen patients and potential research participants not just for Aβ, but for phosphorylated tau, too, he added. "Just looking at Aβ alone is insufficient—you have to look at what's going on with tau," Desikan said.

Fagan agreed, noting that there is plenty of evidence already that tau and Aβ42 together play a role in the cognitive decline of MCI and AD (see Hansson et al., 2006, ARF related news story, and ARF news story). Plus, cross-sectional studies of biomarkers suggest that CSF Aβ levels start dropping as many as 15 years before the onset of symptoms, while tau levels start to rise only five years prior to disease (see also Li et al., 2007, as well as Sunderland et al., 2003). Neither biomarker is predictive on its own, but there is a window of time just prior to disease onset where tau seems to be high and Aβ low in the CSF. "You have to have both pathologies in order for it to be clinically meaningful," Fagan said, referencing cross-sectional data to date. Now, more longitudinal data are needed to probe the relationship between tau and Aβ42, she added. "I think the relationship between them is critical and somehow needs to be defined in a quantitative way."

Ewers and colleagues did report in their paper that the high-Aβ MCI group had elevated CSF phosphorylated tau levels compared to healthy controls, with a similarly high level of CSF Aβ1-42. However, the team did not test phosphorylated tau as a predictor in the current study, so they can draw no conclusions about its role, Ewers said. "It's an interesting lead to follow up," he told ARF.—Gwyneth Dickey Zakaib


  1. A very informative study. One question is, Does the phosphorylation status of CSF tau reflect that of intraneuronal tau?

    View all comments by Takaomi Saido
  2. Ewers et al. use data from ADNI to understand relationships between β amyloid (Aβ) and longitudinal change in brain structure and function (measured with MRI and FDG-PET, respectively), as well as longitudinal cognitive decline. The results provide an in-depth view of complex relationships, revealing robust effects in MCI—higher Aβ was associated with volume loss across many regions of interest (ROIs) and decline in metabolism, as well as longitudinal change in cognition—and weak to null relationships within normals (a relationship was only seen between Aβ and volume loss in a few ROIs), despite comparable levels of Aβ in Aβ-positive MCI and normal subjects.

    These analyses reiterate an important question that resonates in studies unveiling elevated Aβ in cognitively normal elderly controls: How is it possible that these subjects with high pathological burden remain normal? Although this question is not directly tested by Ewers et al., the weaker relationships between amyloid and brain measures in Aβ-positive normals compared to Aβ-positive MCI subjects provide evidence that certain individuals can have high levels of this pathology without concurrent downstream consequences (such as atrophy and hypometabolism).

    There are a number of potential explanations for this discrepancy in Aβ-positive normals. An obvious and simple explanation may be that they have been accumulating Aβ for a shorter amount of time (they are earlier on the trajectory towards AD, before downstream events have occurred), or perhaps have been depositing Aβ at a slower rate (a rate that may not be neurotoxic and/or allows sufficient time for compensatory responses to take place). Another potential mechanism could be via heightened resistance to Aβ pathology in these individuals (perhaps through genetics, neuronal plasticity, high brain reserve, etc.). Finally, it is possible that Aβ-positive normals lack pathologies that exacerbate effects of Aβ (such as neurofibrillary tangle pathology, which may have an etiology independent of Aβ, or may be downstream of Aβ and has not yet emerged in these normals. See Desikan et al., 2011, for an analysis relevant to this claim.) These proposed mechanisms are not necessarily exclusive, and may interact together to enable certain normal individuals to "cope" with Aβ.

    Future studies that address these potential mechanisms will clarify the relevance of this pathology in cognitively normal elderly individuals and during early AD development.


    . Amyloid-β associated volume loss occurs only in the presence of phospho-tau. Ann Neurol. 2011 Oct;70(4):657-61. PubMed.

    View all comments by Elizabeth Mormino
  3. This paper is a good example of the field taking the biomarker issue to the next level—beyond the simple uni-modal and uni-analyte investigations that have been the initial approaches. This research is important not only for identifying biomarkers that have the most clinical potential, but also for better elucidating the pathological processes that are taking place during the natural course of the disease. Now that multi-modal assessments are being implemented in longitudinal studies, I expect we'll see a lot of similar types of analyses. I'm not sure what to make of the positive findings for phosphorylated tau, but not total tau, since these two markers are very highly correlated in AD, but they may suggest some interesting things specifically related to tau hyperphosphorylation. The overall findings of Desikan et al. clearly deserve attention using a different independent cohort.

    View all comments by Anne Fagan

Make a Comment

To make a comment you must login or register.


Webinar Citations

  1. Evolution of AD Trials

News Citations

  1. Miami: Is Human Amyloid Imaging Ready for Clinical Trials?
  2. Honolulu: The Missing Link? Tau Mediates Aβ Toxicity at Synapse
  3. APP Mice: Losing Tau Solves Their Memory Problems
  4. Biomarker Roundup: Collecting Clues from MRIs to RNAs
  5. Diagnosis of AD—Does Spinal Fluid Hold the Key?

Paper Citations

  1. . Clinico-pathologic studies in dementia: nondemented subjects with pathologically confirmed Alzheimer's disease. Neurology. 1988 Nov;38(11):1682-7. PubMed.
  2. . Dendritic function of tau mediates amyloid-beta toxicity in Alzheimer's disease mouse models. Cell. 2010 Aug 6;142(3):387-97. Epub 2010 Jul 22 PubMed.
  3. . Reducing endogenous tau ameliorates amyloid beta-induced deficits in an Alzheimer's disease mouse model. Science. 2007 May 4;316(5825):750-4. PubMed.
  4. . Association between CSF biomarkers and incipient Alzheimer's disease in patients with mild cognitive impairment: a follow-up study. Lancet Neurol. 2006 Mar;5(3):228-34. PubMed.
  5. . CSF tau/Abeta42 ratio for increased risk of mild cognitive impairment: a follow-up study. Neurology. 2007 Aug 14;69(7):631-9. PubMed.
  6. . Decreased beta-amyloid1-42 and increased tau levels in cerebrospinal fluid of patients with Alzheimer disease. JAMA. 2003 Apr 23-30;289(16):2094-103. PubMed.

External Citations

  1. Alzheimer’s Disease Neuroimaging Initiative

Further Reading

Primary Papers

  1. . Amyloid-β associated volume loss occurs only in the presence of phospho-tau. Ann Neurol. 2011 Oct;70(4):657-61. PubMed.
  2. . CSF Biomarker and PIB-PET-Derived Beta-Amyloid Signature Predicts Metabolic, Gray Matter, and Cognitive Changes in Nondemented Subjects. Cereb Cortex. 2011 Oct 29; PubMed.